Mapping the EORTC QLQ-C30 and QLQ-H&N35 to the EQ-5D for head and neck cancer: Can disease-specific utilities be obtained?

• Published in: PloS one Vol. 14; no. 12; p. e0226077
• Main Authors: Beck, Ann-Jean C. C., Kieffer, Jacobien M., Retèl, Valesca P., van Overveld, Lydia F. J., Takes, Robert P., van den Brekel, Michiel W. M., van Harten, Wim H., Stuiver, Martijn M.
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 13.12.2019
• Subjects: Adult; Analysis; Antineoplastic Agents - therapeutic use; Bayes Theorem; Bayesian analysis; Biology and Life Sciences; Cancer; Cancer treatment; Care and treatment; Cost analysis; Cost benefit analysis; Economics; Emotions; Epidemiology; Female; Global health; Head & neck cancer; Head and neck cancer; Head and Neck Neoplasms - drug therapy; Head and Neck Neoplasms - pathology; Head and Neck Neoplasms - psychology; Health sciences; Health Status; Hospitals; Humans; Innovations; Least-Squares Analysis; Male; Mapping; Maxillofacial surgery; Medical equipment; Medicine and Health Sciences; Middle Aged; Models, Statistical; Oncology; Pain; Patients; Performance evaluation; Quality of Life; Questionnaires; Radiation therapy; Regression analysis; Regression models; Research and Analysis Methods; Shrinkage; Social Sciences; Statistical analysis; Surgery; Systematic review; Technological change; Utilities; World health; United Kingdom > UK; England; Netherlands
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0226077

Innovations in head and neck cancer (HNC) treatment are often subject to economic evaluation prior to their reimbursement and subsequent access for patients. Mapping functions facilitate economic evaluation of new treatments when the required utility data is absent, but quality of life data is available. The objective of this study is to develop a mapping function translating the EORTC QLQ-C30 to EQ-5D-derived utilities for HNC through regression modeling, and to explore the added value of disease-specific EORTC QLQ-H&N35 scales to the model. Data was obtained on patients with primary HNC treated with curative intent derived from two hospitals. Model development was conducted in two phases: 1. Predictor selection based on theory- and data-driven methods, resulting in three sets of potential predictors from the quality of life questionnaires; 2. Selection of the best out of four methods: ordinary-least squares, mixed-effects linear, Cox and beta regression, using the first set of predictors from EORTC QLQ-C30 scales with most correspondence to EQ-5D dimensions. Using a stepwise approach, we assessed added values of predictors in the other two sets. Model fit was assessed using Akaike and Bayesian Information Criterion (AIC and BIC) and model performance was evaluated by MAE, RMSE and limits of agreement (LOA). The beta regression model showed best model fit, with global health status, physical-, role- and emotional functioning and pain scales as predictors. Adding HNC-specific scales did not improve the model. Model performance was reasonable; R2 = 0.39, MAE = 0.0949, RMSE = 0.1209, 95% LOA of -0.243 to 0.231 (bias -0.01), with an error correlation of 0.32. The estimated shrinkage factor was 0.90. Selected scales from the EORTC QLQ-C30 can be used to estimate utilities for HNC using beta regression. Including EORTC QLQ-H&N35 scales does not improve the mapping function. The mapping model may serve as a tool to enable cost-effectiveness analyses of innovative HNC treatments, for example for reimbursement issues. Further research should assess the robustness and generalizability of the function by validating the model in an external cohort of HNC patients.